Monte Carlo Sampling Based Collision Detection Algorithm Development And False Positive And False Negative Rate Analysis: A Bayesian Approach

Period of Performance: 06/02/2010 - 06/01/2011

$747K

Phase 2 STTR

Recipient Firm

Princeton Vision LLC
5 Banff Dr.
Princeton Junctio, NJ 08550
Principal Investigator
Firm POC

Research Institution

Carnegie Mellon University
5000 Forbes Avenue
Pittsburgh, PA 15213
Institution POC

Abstract

In this Phase II proposal, the main thrust is to build a hardware MCICD prototype, and validate the FAR/FNR through real vehicle testing. By leveraging the existing LADAR based sensing platform in CMU, we expect to shorten the development cycle and reduce the overall cost. Extensive real vehicle testing is expected both in staged scenarios and in normal traffic. In this Phase II program, we also propose to further improve the MCICD algorithm. The tasks include: (1) further improve the LADAR sensor noise model, with special attention on the correlation effect, (2) study and develop the correlated sampling method, and (3) study and improve the sampling efficiency. During the Phase II program, we expect to have a extensively tested MCICD system in a relatively compact package. We plan to promote the MCICD technology through the marketing channels. Our initial targeting market is the UGV or military patrol vehicle protection, and we are optimistic on the market potential.